SPLITTING LIGHT VERBS IN THE RESULTATIVE CONSTRUCTION

SPLITTING LIGHT VERBS IN THE RESULTATIVE CONSTRUCTION

by

© Yanxiao Ma A Thesis submitted to the School of Graduate Studies

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy Depar tment of Linguistics Memorial University of Newfoundland

May, 2019

St. John’s Newfoundland and Labrador

ABSTRACT

This thesis examines the resultative construction in Mandarin Chinese, and extends to different resultative patterns in Shanghai dialect and Innu-aimûn. The objective of the study is to investigate the derivational process of various resultative patterns with the labeling theory. I will attempt to provide a symmetric analysis of the resultative construction in general. Hopefully, this study will shed some light on the interface study between the syntactic derivation and the event semantic analysis in the resultative construction.

In Mandarin Chinese, a v-splitting structure is proposed in compounding resultatives and DE-resultatives, in which multiple adjacent light verbs are hypothesized along with feature inheritance. The v-splitting structure is also applied to a pattern containing a preverbal resultative adverb in Mandarin Chinese, which is termed as adverb-resultatives. I propose that these patterns fall into the resultative construction in a broad sense. The splitting approach is better than a base-generated structure with two v heads. One advantage is that it allows us to formulate the analysis in which the root raises to v* without violating the head movement constraint. It also works better to explain the specificity effect, when it is based on a splitting structure and labeling requirement.

Extending the study to the cross-linguistic scope, resultatives in Shanghai dialect and Innu-

aimûn are briefly explored. On the one hand, a mono-layer light verb is proposed in

Shanghai dialect, in which the resultative predicate does not undergo head movement to

the light verb, ending up with the serial verb pattern. The contrast between Shanghai and Mandarin resultatives indicates the diachronical development of light verbs in Chinese:

from a unified mono-layer to a v-splitting structure. On the other hand, in Innu-aimûn, the resultative predicate is realized in the preverbal position within the complex verb structure.

Multiple head movement is analogically explored in Innu-aimûn.

Through the comparative study, the parameter of head movement is emphasized in various resultative patterns. Different strategies to the symmetry-breaking are proposed across languages. In English, the <φ,φ> label matching is used in the VP agreement system.

However, this is not unanimously applied to other languages. In Innu-aimûn, result-raising

in the resultative construction is the only possibility to break the symmetry of an

uninterpretable configuration. In Chinese, two ways are proposed on the symmetry-

breaking: the <φ,φ> feature-matching as the core mechanism in the VP agreement system,

and result-raising as a supplementary operation.

ACKNOWLEDGEMENTS

The acknowledgement page provides me a precious opportunity to express my sincere gratitude to those who have offered me constant support for my overseas study and the thesis completion. My heartfelt gratitude first goes to my thesis committee examiners Dr.

Keir Moulton, Dr. Alexander Williams and Dr. Julie Brittain. Without their helpful comments and advice, the thesis would not have been completed as it is.

My deepest indebtedness doubtless to my supervisor Dr. Phil Branigan. Indeed, I am so honored to be a graduate student working with Phil, and it is Phil that makes my overseas study enriched. Even up to now when I type the acknowledgement words, what came into my mind is that Phil is always reachable and ready to listen to my confusions in the thesis writing, regardless of semester breaks or holidays. I have benefited a lot from his patient discussions, insightful comments and endless encouragement. I am also impressed by his insightful sharing in our syntax class and reading groups. The reading group provides good moments for us to read and discuss the frontier syntactic work. Being my supervisor, Phil’s professional expertise and easygoing personality will benefit me for good in my future academic or non-academic life.

In addition, please allow me to speak out my sincere gratitude to Dr. Paul DeDecker, who

advises my first comprehensive examination. I am appreciated that Paul is always kind and

patient to demonstrate useful methods on data-recording and statistic-analysing. I am also

obliged to Dr. Douglas Wharram. I have worked as a teaching assistant in his semantic class. Sitting in his class, his professional expertise largely broadens my linguistic horizon on semantics. I am also so honored to have him as my teaching mentor in the Teaching Skills Enhancement Program. He helps me record my teaching practice and offers many useful comments and advice for my teaching. I also want to say thanks to Dr. Julie Britain.

She is always considerate and gives us limitless support in and out of her class. I really benefit a lot from her teaching and great encouragement. I also wish my immeasurable thanks go to Dr. Sara Mackenzie and Dr. Yvan Rose. Sitting in the phonetic and phonological classes, I am always deeply impressed by their professional expertise and the comfortable class atmosphere. I am so fortune to meet them during my overseas graduate study.

I sincerely hope my thankfulness extends to my dear peer colleagues in the department of Linguistics: Mariana, Alyson, Ismar, Pooching, Annaer, Leilla, Saleem, Hussein Hamid, Kang, Yi, and many others. I appreciate they are always reachable for my thesis and discussion. Their precious comments and fruitful advice help me more than imagined.

Next, please allow me to emphasize the limitless support from Dr. Fuzhen Si, the supervisor

for my master degree. It is Dr. Si that opens the window of linguistics for me, and I am

very honored to be one of her students. I would like to thank her endless encouragement

for my academic life. Without her, I may not have much courage to pursue the overseas

study. Even though staying far away from home, I am obliged that she is always there not only for my study but also for my life advising.

Last but not least, I would like to show my gratitude to my families and friends. The

countless support from them adds outstanding scenery in my overseas life. My parents

always give me unconditional support for my study and many other choices. During the

past oversea years, I cannot go back to visit them as usual. But I feel so excited when we

have the video or audio online call each time. In order not to distract my attention on my

thesis writing and my overseas study, they always share happy experiences with me, and

never mention any difficulties they come across. I know this is all about parents’ love. Here

I would also like to say thanks to my older brother and my sister in-law. They try their best

to take good care of my parents which I should have done. At last, please allow me to

mention a little angel girl, my 4-year-old niece. Although now she cannot understand the

English words here, but I will not hesitate to let her know that I love her, as she has created

immeasurable happy moments for my overseas study.

TABLE OF CONTENTS

CHAPTER ONE INTRODUCTION ... 1

1. THE PHENOMENA ... 2

2. BACKGROUND ASSUMPTIONS ... 4

2.1 H EAD MOVEMENT ... 4

2.1.1 VP shell structure ... 4

2.1.2 Verbal head-movement in Mandarin Chinese ... 8

2.2 L ABELING A LGORITHM ... 14

2.2.1 Labeling rules ... 14

2.2.2 Illustration in verb copying sentences ... 16

3. OVERALL ARGUMENT ... 19

4. LAYOUT OF THESIS ... 20

CHAPTER TWO LITERATURE REVIEW ... 23

1. RESULTATIVES IN GENERATIVE GRAMMAR ... 23

1.1 E VENT STRUCTURES IN RESULTATIVES ... 25

1.2 S EMANTIC ACCOUNT OF ARGUMENT REALIZATION ... 28

1.3 U NDERLYING SYNTACTIC REPRESENTATIONS ... 31

1.3.1 Small clause analyses ... 31

1.3.2 Ternary branching analyses ... 33

1.4 R ESULTATIVE ADVERBS IN GENERATIVE GRAMMAR ... 35

2. RESULTATIVES IN MANDARIN CHINESE ... 39

2.1 L ITERATURE ON C OMPOUNDING R ESULTATIVES ... 39

2.1.1 Directionality on headness ... 39

2.1.2 Lexical Rules ... 42

2.1.3 Small clause analysis ... 45

2.1.4 vP shell structure ... 46

2.1.5 Specificity effects in compounding resultatives ... 47

2.2 L ITERATURE ON DE- RESULTATIVES ... 51

2.2.1 Small clause analysis ... 52

2.2.2 vP shell analysis ... 53

2.2.3 Specificity effects in DE-resultatives ... 55

2.3. A DVERB RESULTATIVES IN M ANDARIN ... 57

2.3.1 Distribution of adverbs ... 57

2.3.2 Predicate analysis ... 60

3. TYPOLOGIES OF CHINESE RESULTATIVES ... 62

4. SUMMARY ... 65

CHAPTER THREE TYPICAL RESULTATIVES ... 67

1. A QUICK SUMMARY OF THE MODEL ... 68

1.1 V - SPLITTING HYPOTHESIS ... 71

1.2 H EAD MOVEMENT CONSTRAINT WITH RECONCILIATION ... 73

1.3 L ANDING SITE OF A RAISED OBJECT ... 75

2. V-SPLITTING IN DE-RESULTATIVES ... 78

2.1 V - SPLITTING DERIVATION IN DE- RESULTATIVES ... 79

2.2 S PECIFICITY IN DE- RESULTATIVES ... 85

2.3 C OMPARISON WITH OTHER APPROACHES ... 87

2.3.1 Comparison with other light verb argument ... 88

2.3.2 Comparison with other derivational approaches ... 91

2.4 S UMMARY OF DE- RESULTATIVES ... 96

3. V-SPLITTING IN COMPOUNDING RESULTATIVES ... 97

3.1 V - SPLITTING DERIVATION IN COMPOUNDING RESULTATIVES ... 98

3.2 S PECIFICITY IN COMPOUNDING RESULTATIVES ... 101

3.3 C OMPARISON WITH OTHER APPROACHES ... 105

3.4 P ARAMETRIC VARIATION ON DIRECT OBJECT RESTRICTION ... 109

4. SUMMARY ... 117

CHAPTER FOUR ADVERB RESULTATIVES ... 118

1. ASYMMETRIC ANALYSES FOR MANNER AND RESULTATIVE ADVERBS ... 120

1.1 M ANNER ADVERBS AS MODIFICATION ... 120

1.2 R ESULTATIVE ADVERBS AS PREDICATION ... 124

2. LABELING ANALYSIS OF ADVERB RESULTATIVES ... 127

2.1 A DVERB RESULTATIVES IN C HINESE ... 129

2.2 S PECIFICITY IN ADVERB RESULTATIVES ... 133

2.3 I NTERVENTION EFFECT IN ADVERB RESULTATIVES ... 135

2.4 C ONTRAST TO E NGLISH A DVERB RESULTATIVES ... 138

3. ADVERB CONJUNCTION ... 140

3.1 A DVERB DISTRIBUTION IN CONJUNCTION ... 142

3.1.1 Flexible orders in typical conjunction ... 142

3.1.2 Asymmetric orders in pseudo-conjunction ... 144

3.1.3 Ellipsis of and-conjunction ... 147

3.2 R EDUPLICATIVE CONSTRAINTS IN ADVERB CONJUNCTION ... 149

4. COMPARISON WITH OTHER APPROACHES ... 152

5. SUMMARY ... 155

CHAPTER FIVE CROSS-LINGUISTIC STUDY ON RESULTATIVE PATTERNS

... 157

1. SERIAL VERB RESULTATIVES IN SHANGHAI DIALECT ... 157

1.1 D ISTINCTIONS IN SYNTACTIC BEHAVIORS ... 158

1.2 L ABELING DERIVATION ... 166

1.2.1 Analysis in detail ... 166

1.2.2 Cross-linguistic evidence on morphological merger ... 168

1.3 E XPLANATION ON SYNTACTIC DISTINCTIONS ... 175

1.4 F EATURE - INHERITANCE EXUBERANCE ... 180

2. RESULTATIVES IN INNU-AIMÛN ... 183

2.1 L ABELING DERIVATION ... 184

2.2 S TRATEGIES OF SYMMETRY - BREAKING ... 189

3. SUMMARY ... 195

CHAPTER SIX CONCLUDING REMARKS ... 197

1. CONCLUSION AND IMPLICATION ... 197

2. FUTURE RESEARCH ... 200

ABBREVIATIONS

1, 2, 3 1

, 2

, 3

Person

Adj Adjective

Adv Adverb

AP Adjectival Phrase Asp Aspect

Aux Auxiliary BecP Become Phrase CausP Causative Phrase CL Classifier

CP Complementizer Phrase DOR Direct Object Restriction DP Determiner Phrase EF Edge Feature E-language External Language EP Event Phrase ExtP Extent Phrase FinP Finiteness Phrase FocP Focus Phrase ForceP Force Phrase GENI Genitive H Head

HMC Head Movement Constraint I-language Internal Language

INFL Inflectional Head

Intrans Intransitive Verb

LA Labeling Algorithm

NEG Negation

NP Noun Phrase

OAsp Outer Aspect

Obj Object

P Preposition

p Functional Preposition

PP Preposition Phrase

PRF Functional Particle

PRO Empty Pronoun

PrP Predicate Phrase

R Root

Res Resultative Predicate

RVC Resultative Verb Compounds

SC Small Clause SPEC Specifier Subj Subject TopP Topic Phrase TP Tense Phrase Trans Transitive Verb VP Verb Phrase

vP Functional Light Verb Phrase

V-V Compounding verbs

V1 First verb in compounding verbs

V2 Second verb in compounding verbs

XP/YP/ZP Maximum Projection/Phrase

φ φ-features

(e.g. Person/Number/Gender/Case, etc.)

Chapter 1 Introduction

This thesis focuses on the syntactic derivation of the resultative construction in Mandarin Chinese. In the first chapter, the general phenomena and three goals of this study are first introduced. Then technical assumptions are presented, which matter for the overall investigation. The purpose in introducing these technical assumptions is not to recapitulate minimalist theory, but simply to establish discoveries I will rely on. In the discussion of the technical background, three specific assumptions are illustrated: the structures of vP &

AspP, the process of head-movement within vP and beyond, and the labeling algorithm. I

will first illustrate how it all works in analysis of the verb copying construction, and it will

be further adopted to analyse the resultative construction in this study. Verb movement in

Chinese is generally limited to the VP shell structure (Larson 1988), and does not further

move to a higher functional category, according to Huang’s (1991, 1992, 1994) series of

work (Paul 2000). A functional category Asp is introduced, which may accommodate some

verb movement outside the vP boundary. The syntactic framework illustrated in this chapter

will be applied to analyse multiple resultative patterns in general, and derivational

mechanisms are drawn from the labeling theory (Chomsky 2013, 2014).

1. The Phenomena

The resultative construction is usually composed of two predicates: a means predicate and a result state predicate (Williams 2008). Two resultative patterns are widely acknowledged in Mandarin Chinese: compounding resultatives (1a) and DE-resultatives (1b).

(1) a. Wusong da si le laohu. (Compounding resultatives) Wusong beat die ASP tiger

‘Wusong beat the tiger so that it died.’

b. Wusong da de laohu si le. (DE-resultatives) Wusong beat DE tiger die PRT

‘Wusong beat the tiger so that it died.’ (Zhang 2001: 192)

In (1a), the resultant state si ‘die’ immediately follows the activity verb da ‘beat’, which is known as the compounding resultative pattern. In (1b), the activity verb da ‘beat’ is followed by the DE-marker, introducing the resultant state si ‘die’. This pattern is known as the DE-resultative pattern.

In addition to these two resultative patterns, the resultant state can be realized as an adverb, immediately preceding the activity verb. As in (2), the resultant state cuicuide ‘crisp’ is realized as an adverb in Mandarin Chinese, denoting the change of the state by completing the activity zha ‘fry’. This pattern will be termed as the adverb resultative pattern in this thesis.

(2) Ta cuicuide zha le yi pan huashengmi.

He crisp fry ASP one plate peanut

‘He fried a plate of peanuts, as a result the peanuts are crisp.’ (Lu 2003)

There is abundant literature to discuss the compounding and the DE-resultative pattern in Chinese. However, few studies (Xiong 2013) put the adverb pattern into the resultative construction in general. I hope this thesis can reach three goals. First, I will describe the resultative data in Mandarin Chinese (i.e. the compounding resultative pattern, the DE- resultative pattern, and the adverb resultative pattern). Second, existing accounts do not seem to well characterize the data as they should. They do not provide a unified explanation in Mandarin Chinese. I will explain these three patterns by providing a unified discussion on their syntactic derivational model. Third, very few existing studies have provided good explanations on derivational differences between Mandarin Chinese and other Chinese dialects or other languages. Motivated by these general problems, this thesis focuses on the syntactic derivation of three patterns in Mandarin Chinese, and then extend to Shanghai dialect and other languages, like English and Innu-aimûn. The resultative construction exists in these languages, but the resultant states are realized in different ways. I will explain the resultative variety from a v-splitting approach, hoping to provide a unified explanation to the resultative construction in general.

In the resultative construction, the syntactic relation between the activity verb and the

resultant predicate will be explained, based on the vP shell structure. The understanding of

the vP structure is fundamental to analyse head movement and to specify the internal

structure of the resultative construction. The background assumption on the vP shell

structure will be presented in the following.

2. Background Assumptions

After introducing the investigated data and existing problems, I will provide background theory in this section. Three areas are illustrated: the structures of vP & AspP, the process of head-movement within vP and beyond, and the labeling algorithm. The theoretical assumptions will be later adopted to analyse the resultative construction in general.

2.1 Head movement 2.1.1 VP shell structure

Head movement within the vP structure will be important for my proposal in chapter 3, so

the technical background in this area is necessary to be introduced. Shell theory is first

developed to analyse the double object construction in English (Larson 1998), in which the

indirect object and the verb merge as a constituent, and the direct object is derived as the

subject of the verbal phrase. As in (3), the verb is raised to a higher functional verbal

category, via the V-raising operation (Larson 1988).

(3) a. John sent a letter to Mary.

b.

(Larson 1988: 343)

A similar verb raising operation is found in passive sentences in (4). The object Mary

becomes the subject in Spec-vP, which is available to the higher T and move further to

Spec-TP (Larson 2014). The verb is raised to a higher functional verbal category through

V-raising.

(4) a. Mary was kissed by John.

b.

(Larson 2014: 99)

Next, let us look at the syntactic nature of the verb root in particular. Marantz (1997) discusses the role of roots playing for (Phonological) Words from a syntactic perspective, against lexicalist assumption. He argues that roots are category neutral and are defined as things with specific meaning. The full meaning of the root is determined by functional categories attached to the root in a special boundary context. The boundary context provides particular syntactic environment, in which the categoryless root is created a verbalizing category by the functional head v, or a nominalizing category by the functional D. Marantz classifies Roots into three types, based on Chomsky’s (1970) remarks on nominalization: roots not internally caused, roots internally caused, and roots naming an ending result. The first two classes denote changes of the state, and the third type names an ending result, not an event of changing state. As in (5), an external agent is compatible with

Mary v’

v v kiss

VP

PP V’

by John

kiss Mary

vP

DESTROY-type roots, but not in GROW-type roots. The external agent is still implied in GROW-like roots, but only in verbal environment. Since no v-1 agent is interpreted in the internally caused change of state. Chomsky solves this paradox by assuming that the agent of grow is restricted to projection in verbal environment, but the agent of destroy (externally caused change of state) is interpreted by the root. BREAK-type roots license an optional external agent. The verbal categories are implied by “v”, and the nominal use of the root is derived by merging a root with “D”.

(5) Three Types of Roots

Root Class

DESTROY change of state, not internally caused

GROW change of state, internally caused

BREAK result (of change of state)

(Marantz 1997: 217)

Let us see the instantiation of the root DESTROY in detail in (6), which projects an external

agent. The external agent is implied by v-1, projecting a verbalizing context, thus a verbal

category is derived in (6a). The nominal context of the DESTROY root is derived by

merging it with “D” in (6b). An external agent can also be found in the nominalized

destruction, but the agentive interpretation for the possessor DP is not introduced by v-1,

but rather implied by an externally caused event. The purpose is to distinguish destruction

and growth, which allows an external agent in the former but not in the latter.

However, the -ing nominalized form derives first from a verbalizing and then a nominalizing context.

The -ing nominalized process is represented in (6c).

(6) Nominalization Derived from Verb

v-1

(Marantz 1997) I will adopt Marantz’s conclusion and assume that Root is category-neutral when it enters the derivation. The category of Root is labeled by a higher functional category.

2.1.2 Verbal head-movement in Mandarin Chinese

Now let us move forward to head raising in Chinese. Based on Larson’s (1988) vP shell analysis, Huang (1991) argues that the topmost empty verb in the VP shell structure is the

1

If v-1 is introduced in the agentive interpretation in the footnote (i)-a, we would also expect the same reading in (i)-b, since the agent-introducing v-1 should also be allowed within the nominalization (Marantz 1997: 218).

(i) a. John’s destruction of the city.

b.*John’s growth of tomatoes.

v-1 ÖDESTROY ÖDESTROY the city

a. John destroyed the city.

host of an eventuality predicate in Chinese. The eventuality predicate denotes the event type of the characterized verb. For instance, the eventuality predicate DO is indicated in activity verbs, BE/HOLD in stative verbs, and CAUSE in causative verbs. It is the eventuality predicate that triggers head movement in Chinese, although it is phonetically null. In Huang’s work, the landing site of the raised verb in Chinese is confined to the vP shell structure. As we can see in (7b), the verb zui ‘get drunk’ (along with the affixed de) is raised to CAUSE, which is still confined into the vP shell structure.

(7) a. Na ping jiu zui de Zhangsan zhan bu qilai.

that CL wine drunk DE Zhangsan stand NEG rise

‘That bottle of wine got Zhangsan so drunk that he could not stand up.’

b. Na Ping jiu [

CAUSE

[

Zhangsan zui

de zhan bu qilai]]

that CL wine Zhangsan drunk DE stand NEG rise (Huang 1992) Paul (2000) argues that head movement in Chinese may go beyond the vP shell structure, rather than always being confined into the verb phrase. Supporting evidence is observed in verb copying sentences, in which verbs in the same form are used twice.

(8) a. Wo qi ma qi le haojici le.

I ride horse ride ASP many-CL PRF ‘I have ridden on a horse many times.’

b. Wo qi ma yijing qi le haoji ci le.

I ride horse already ride ASP many CL PRF

‘I have ridden on a horse already many times.’ (Paul 2000: 266)

In (8a), the verb qi ‘ride’ has been used twice. In the verb copying construction, the adverb

yijing ‘already’ occurs right before the second V2, as in (8b). Paul (2000) explains that

adverbs enter the derivation in the edge of vP. The second verb V2 occupies within vP, but

V1 must be outside vP. Based on the adverb distribution in verb copying sentences, it at least indicates that head movement in Chinese is not always confined to the vP shell structure. The discussion on head movement in Chinese is more complex than the general vP shell structure, and it needs reconsideration (Paul 2000).

Based on Paul’s (2000) finding that the landing site of a raised verb may go beyond the vP shell structure in Chinese, I suppose a functional head Asp(ect) outside the phasal v*P. The interpretable [ASPECTUAL] ([ASP]) feature is originated in Asp, which denotes the event type of the predicate. The [ASP] feature in Asp can be either strong or weak. If the [ASP]

is strong, then it can trigger raising of a verb. If it is weak, no head is raised to Asp. The derivational structure is presented in (9).

(9) vP-external Asp

Asp

v*

Spec

Root DP [EVENT]

optional

The idea on optional feature-driven head movement is supported by Pollock’s (1989) analysis in French. In French, lexical verbs (and auxiliaries) undergo verb movement in tensed clauses in (10a), but not necessary in infinite clauses in (10b-10c).

(10) a. Jean (n’) aime pas Marie.

John like not Mary ‘John does not like Mary.’

b. N’avoir/*posséder pas de voiture en banlieue crée des problèmes.

‘To have/possess not a car in the suburbs creates problems.’

c. (Souvent) Paraître (souvent) triste pendant son voyage de noce, c’est rare.

‘(Often) to-appear (often) sad during one’s honeymoon is rare.’ (Pollock 1989)

According to Pollock (1989), the finite lexical verb aime in (10a) obligatorily moves to Infl (i.e. T), triggered by [+finite] feature in Infl. Main verb infinites cannot raise over negation, but auxiliaries are allowed, in (10b). Although main verb infinites cannot move over negation, it can optionally precede VP-adverbs (or floated quantifiers) in (10c). Pollock (1989) assumes the split-Infl structure to analyse the finite and infinite contrast.

(11)

(Pollock 1989)

In (11), main verbs in finite clauses obligatorily move to I1 (the higher Infl); main verbs in infinitives optionally raise to I2 (the lower Infl). The head movement distinction is correlated to [+/-finite] features and [v] features in Infl. In finite clauses, the [+finite]

feature always triggers the main verb to Infl. In infinite clauses, the strong [v] feature in Infl raises main verbs to the lower Infl. Otherwise, if the [v] feature is weak, no head movement is observed. Thus, optional head movement is tested in French, which is correlated to feature-driven motivation.

So far, I have proposed that the verb may optionally raise to Asp in Chinese, if a strong [ASP] feature is contained in Asp. This proposal is comparable to French optional head movement (Pollock 1989).

The structure of vP is connected to the range of the position where adverbs may appear.

Let us move from head raising to the syntactic position of adverbs in the derivation.

Following Bowers (1993), manner adverbs enter the derivation in specifiers of an intermediate projection. The intermediate position is the intermediate merger result of a head and its complement. Paul (2000) assumes the edge of vP to host manner adverbs in Mandarin Chinese (as I have mentioned in the verb copying sentences). Thus, the Spec-v*

is the possible position (represented by number 2) to accommodate manner adverbs, but

not the Spec-Asp (represented by number 1)

and the Spec-Root (represented by number 3) in Chinese.

(12)

In Chinese, manner adverbs are allowed in preverbal positions, but not in sentence-final positions in a context-independent structure. However, the adverb distribution may vary across languages in the surface order, due to the direction of adverb adjunction (Bowers 1993). For instance, adverbs can adjoin either to the left or to the right of a specifier position in English, so both preverbal and sentence-final adverbs are observed. The empirical application of Asp outside v*P and the adverb in the Spec-v*P will be adopted in chapter 4 to explain the derivation of Chinese adverb resultatives in detail.

2

I will show the adverb can be contained in the Aspectual phrase only in pseudo-conjunction in Chinese.

Asp

v*

Spec

Root DP Spec

③ ②

①

2.2 Labeling Algorithm 2.2.1 Labeling rules

Now, let us move on to the labeling algorithm, which matters to describe the resultative head moving to a higher position in the resultative construction. In this section, I will introduce this framework in general. Chomsky (2013) argues that the label of the syntactic object is dependent on the computation, into which each of the syntactic objects enters.

Following Marantz (1997), any category is created by a functional head in a particular boundary context. For instance, Root is labeled as a verbal category by the functional v, and as a nominal category by a higher D. Beyond this, Chomsky (2013: 43) claims labeling is essentially minimally search. In the configuration [H, XP], with H being a phasal head, the Labeling Algorithm (LA) takes H as the label. For instance, in {v, {V, DP}}, the label of {V, DP} must be available before {v, {V, DP}}, and LA takes v as the label. In the configuration [H, XP], with H being a non-phasal head, either the procedure (13a) or (13b) is chosen (Chomsky 2013; Chomsky 2014; cited in Hosono, 2015: 53-54):

(13) a. The subject in [Spec, XP] moves to [Spec, H] and strengthens the non-phasal head;

b. That raised subject and the non-phasal head agree in some feature; LA takes that shared feature as the label.

According to the LA, the non-phasal heads R(oot) and T are weak and cannot be labeled

by themselves. They must be strengthened by movement of the subject into the specifier

position of their complement, and LA takes the <φ,φ> as the label.

In the configuration [XP, YP], two labeling possibilities are available for the labeling algorithm (Chomsky 2013; Chomsky 2014; cited in Hosono, 2015: 54):

(14) a. If either XP or YP undergoes internal merger, LA takes the head of the phrasal object that does not move out as the label;

b. If an agree relation exists between XP and YP, LA takes that shared feature as the label.

For instance, in [C [T [DP [vP…]]]], the labeling algorithm is represented in (15). In the configuration [DP, vP], DP undergoes internal merger, to a higher position, and the head of the vP that does not move out is the label, i.e. vP. In the configuration [T, vP], the subject in [Spec, vP] (i.e. DP) moves to [Spec, T] and the non-phasal head T gets strengthened.

The raised subject DP shares φ-features with the non-phasal head T, so LA takes φ-features as the label, i.e. φP (<φ, φ>). The phasal head C merges into before internal merger of the subject DP to TP, since T inherits the uF from C. The derivation structure in the labeling algorithm is presented in (15).

(15) Labeling in CP

DP

v

Root

DP C

<φ,φ>

T

DP <φ,φ>

(The labeling approach will be used to explain the resultative construction. It will provide a technical solution to movement of the resultative predicate to a higher position, especially relevant to the resultative adverb when it raises up to a higher position in adverb resultatives in chapter 4.)

2.2.2 Illustration in verb copying sentences

After analysing the basic assumption of head raising and the labeling algorithm, let us move back and see how verb copying sentences might be derived within this framework. The purpose of introducing the structure of verb copying sentences here is to simply support the claim that the verb may undergo head movement and raise beyond the vP shell structure in Chinese. The verb can be located in the functional Asp, containing a strong [ASPECTUAL]

feature. Here I repeat the verb copying sentence in (16).

(16) Wo qi ma qi le haojici le.

I ride horse ride ASP many-CL PRF ‘I have ridden on a horse many times.’

The verb root qi ‘ride’ adjoins to the categorizing functional head v*, making it a verbal category. The verb further moves to the Asp, triggered by a strong [ASPECTUAL] feature.

Then the postverbal object ma ‘horse’ merges with the higher copy of the verb through

sideward movement, forming qi ma ‘ride horse’. The sideward movement is a derivation

operation, in which the copy of a syntactic object merges with another syntactic object

independently, and then enter the derivation in a sideward way (Nunes 2001). In sideward

derivation, the higher copy of the verb qi ‘ride’ merges with the object ma ‘horse’, forming

the assembled constituent qi ma ‘ride horse’. The sideward derivation is triggered by a theta-feature checking to the object by the higher copy of the verb. The theta-feature checking satisfies the Last Resort condition on Copy (Cheng 2007). The higher copy of the verb (Root+v*+Asp) does not c-command the lower copy of the verb (Root+v*), so both copies of head raising are remained and spelt-out in verb copying sentences.

Cheng (2007) assumes that the first verb and the bare noun in verb copying sentences behave like compounds, since the aspectual marker -le can never occur after the first verb.

The first verb and the object feed noun incorporation, producing the VP configuration as a whole unit. This is similarly assumed in Hsu (2008). Hsu argues that the highest copy of verb and the object are topicalised as a whole in the sentence-internal position, since the aspectual marker -le is not possible right after the first verb. The lacking of the aspectual marker -le supports the idea that the first verb falls into part of the incorporated or topicalised complex. For the manner adverb yijing ‘already’, it is legitimate to adjoin to the immediate node of v*P but not so high as the immediate projection of Asp phrase, as assumed in Paul (2000). This is also empirically supported by the ungrammaticality of pre- topical adverbs in Chinese,

if the first verb and the object are analysed as a topic-like contstituent as a whole (Hsu 2008).

3

Adverbs are not allowed in a pre-topical position in Chinese.

(ii)a. Zhe ben shu wo hen renzhende du le yi bian.

this CL book I very carefully read ASP one CL ‘I read this book very carefully.’

b.*Hen renzhende zhe ben shu wo du le yi bian.

very carefully this CL book I read ASP one CL

Beyond the Asp projection, T is hypothesized as general and it triggers the subject from Spec-vP to Spec-T. The derivation of the verb copying sentence is presented in (17), in which a vP-external functional head Asp is hypothesized as the landing site of a raised verb.

(17) Verb copying sentences

In the above section, the labeling algorithm, the vP shell structure and the vP-external Asp are illustrated in analysis of the verb copying construction. The vP-framework will be later adopted to account for different resultative patterns in Mandarin Chinese and then extend to other dialects and languages.

Asp

v*

Subject

Root qi

DP haoji-ci qi ma

Adv yijing

*Adv

yijing

3. Overall Argument

In this thesis, I will first analyse resultative patterns in Mandarin Chinese in detail, and then in Shanghai dialect and Innu-aimûn more quickly, to investigate the derivational process of the resultative construction in general. I propose that the phasal head v* splits into multiple adjacent layers along with feature inheritance, producing a v-splitting structure in two typical resultative patterns in Mandarin Chinese: the compounding pattern and the DE- pattern. Then, I will apply the v-splitting structure to explore Chinese adverb resultatives, concluding that adverb resultatives share a similar underlying base with two widely discussed patterns, and they can be generalised as broad resultatives in Mandarin Chinese.

In the Shanghai dialect, one of the southern Chinese dialects, serial verb resultatives are

attested. I analyse this pattern with a mono-layer light verb structure, which is reminiscent

of the resultative pattern in Middle Chinese. The contrast between the serial verb pattern in

Shanghai dialect and the compounding/DE-resultatives in Mandarin indicates diachronic

development of the light verb system: from a mono-layer to a v-splitting structure. The

analogical cross-linguistic study on Innu-aimûn shows that multiple head movement occurs

in the resultative pattern. The syntactic difference in these dialects and languages will be

discussed from the parametric setting of head movement.

4. Layout of Thesis

This thesis consists of 6 Chapters. I present a rough introduction to each chapter in the following.

Chapter 1 provides basic background assumptions, as I have discussed above. In this chapter, the vP shell structure (Larson 1988) is first introduced, based on which I discuss that vP-external Asp may function as a landing site of head raising in Chinese. In addition, the labeling algorithm (Chomsky 2013, 2014) is demonstrated in this chapter. It will be used to explain the underlying movement motivation in resultative patterns across languages.

Chapter 2 considers the pertinent literature on resultative patterns. General studies on resultatives are introduced in terms of event structure, semantics and various approaches to syntactic derivation. After summarizing general approaches on resultatives, I focus the review on resultatives in Mandarin Chinese. First, two patterns are widely discussed in Mandarin: compounding resultatives and DE-resultatives. Second, adverb resultatives in general (e.g. the data in English) are introduced, and then move on to the pattern containing a preverbal resultative adverb in Mandarin Chinese.

Chapter 3 focuses on typical resultatives in Mandarin. Compounding resultatives and DE-

resultatives are analysed from the v-splitting structure. Comparing the compounding

pattern and the DE-resultative pattern, I propose that both are derived within a v-splitting structure. These two patterns share a similar underlying model. In addition, I also explain specificity asymmetries in these two typical resultative patterns from the labeling perspective. The asymmetric (sisterhood) relationship between the postverbal object and the resultative head causes the specificity constraint.

Chapter 4 examines the derivation of adverb resultatives in Mandarin Chinese. A v-splitting structure is tested and successive cyclic head movement Res-to-v-Asp is discussed in this pattern. In this chapter, the adverb conjunction is also explored; typical and pseudo conjunction are distinguished. The conjunction difference is explained by the v-splitting model and with the phonological interface constraint.

Chapter 5 compares resultatives across dialects/languages. A micro-parametric study is first done in Shanghai dialect. Unlike the v-splitting structure in Mandarin, a mono-layer light verb is analysed in Shanghai serial verb resultatives. In addition, a macro-parametric study is taken in Innu-aimûn, an Algonquian language spoken in northeastern Canada.

Multiple head movement is hypothesized in resultatives in Innu-aimûn. The discrepancy of structural characteristics across languages is explained by non-local head movement, the light verb system, and symmetry-breaking in the labeling theory.

Chapter 6 draws a conclusion to this thesis. In this thesis, the labeling-based analyses on a

set of resultative patterns are meant to provide a systematic perspective on how resultatives

work in Mandarin Chinese, Shanghai dialect and Innu-aimûn. However, more cross-

linguistic researches should still be conducted in order to present a better understanding of

the resultative construction in general.

Chapter 2 Literature Review

In this chapter, I will first introduce the resultative literature in generative grammar and then move to studies in Mandarin Chinese. Three general approaches to the resultative construction are introduced in section 1: event structures, semantic analysis of argument realizations and the syntactic derivation of the resultative structure. In this section, I also introduce the generative study of adverb resultatives, mainly based on English data. In section 2, specific studies on Chinese resultatives are presented. Three patterns are introduced in turn: the compounding resultative pattern, the DE-phrasal resultative pattern and the adverb resultative pattern. The classification of Chinese resultatives is introduced in section 3.

1. Resultatives in Generative Grammar

There is abundant literature dealing with resultatives in linguistics, because they occupy an

important place on the interface of lexical semantics and syntax. The resultative

construction consists of two predicates: the first describing an activity predicate, and the

second for a result predicate (Simpson 1983; Carrier & Randall 1992; Rappaport Hovav

and Levin 2001). In following examples in (1), the adjectives flat, hoarse and solid describe

the change of state. The change of state is a result of completing an action, denoted by the

activity verb, such as hammer, yell, and freeze in (1).

(1) a. The blacksmith hammered the metal flat.

b. Sheila yelled herself hoarse.

c. The river froze solid. (Levin & Rappaport Hovav 1995: 39; 2005: 219)

The resultative construction is often in “the form of NP1 V NP2 XP, in which the verb may be a transitive or an unergative intransitive” (Levin 2013: 1). The resultative predicate is fairly free in terms of category, and it may be an adjective phrase, a prepositional phrase, a noun phrase, or a verbal particle, as in (2).

(2) Two groups of resultatives Group I: Transitive resultatives

a. The gardener watered the tulips flat.

b. The grocer ground the coffee beans into a fine powder.

c. They painted the house a hideous shade of green.

Group II: Intransitive resultatives

a. The joggers ran their Nikes threadbare.

b. The kids laughed themselves into a frenzy.

c. He sneezed his handkerchief completely soggy.

(Carrier and Randall 1992)

The word order mainly varies in two ways: the object (direct or not direct) surfacing

between V and Res in a discontinuous order, as in English in (2); or in a continuous order,

with the object following Res, as in Igbo, Ambae in (3). Res may also precede V, like in

German (Williams 2007) in (4) or in Innu-aimûn in (5) (Branigan 2018). The Res-V order

is found not only in head final languages (Williams 2007), but also in polysynthetic

languages (Branigan 2018).

(3) a. Ọkụkọ kpu wa -ra akw’a.

fowl brood split -FACT egg

‘The chicken made the eggs split by brooding.’ [Igbo (Benue-Congo)]

b. da =mo tai visa na avi.

1ns.incS =REAL chop split(tr) ACC firewood

‘We split the firewood by chopping it.’ [Ambae (Oceanic, Vanuatu) (Williams 2007)

(4) a. weil er meine Hosen unheimlich sauber gewaschen hat because he my pants uncannily clean washed has

‘because he made my pants uncannily clean by washing’ [German]

b. Die Teekanne leer trinken the teapot empty drink

‘To drink the teapot empty’ [German]

c. Die Tulpen platt Giessen the tulips flat water

‘Water the tulips flat’ [German] (Kratzer 2005) (5) a. Pien shîpeku-peshaim

mitshuâpinnû.

Pien shîpeku=peshaim

mitshuâp-innû.

Peter green-paint house

‘Peter painted the house green.’

b. Ânî kashteu-peshaim

. Ânî kashteu=peshaim

. Annie black=paint

‘Annie painted it black.’ (Branigan 2018: 208) In the following section, the event structure approach, a semantic account for argument

realization, and underlying syntactic representation will be introduced to show the range of analyses for the resultative construction.

1.1 Event structures in resultatives

First, I will introduce the event structure analysis. Although I will not ultimately adopt the

semantic account, the introduction to this approach provides a typological perspective to

the syntactic analysis in this construction. The lexical semantic representation of verbs

starts from Vendler’s (1957) classification into four event types. States represent static situations, but Activities express dynamic situations. Both States and Activities denote temporally unbounded/atelic situations. Achievements are punctual, while Accomplishments extend over a period time. Both Achievements and Accomplishments express a change of state, so they are temporarily bounded/telic.

(6) Four Event Types of Verbs a. States (static/atelic)

b. Activities (dynamic/atelic) c. Accomplishments (telic)

d. Achievements (telic) (Vendler 1957) Dowty (1979) further proposes three primitives DO, CAUSE, BECOME to express event types of the verb. The primitive DO is assumed in activities, BECOME in achievements, and CAUSE in accomplishments in (7).

(7) Primitives in Event Types a. state: p

(a

a

)

b. activity: DO (a

[p

(a

a

)]) c. achievement: BECOME [p

(a

a

)]

d. accomplishment: [[DO (a

[p

(a

a

)])] CAUSE [BECOME [p

(a

a

)]]]

(Dowty 1979: 123-124)

In states, activities and achievements, a single event is assumed. In accomplishments, the

lexical semantic representation consists of complex events: a causing subevent and a result

subevent.

Following Dowty, Rappaport Hovav and Levin (1998) propose that verb meanings are complex, and verbs directly lexicalize complex event structures, as in (8).

(8) Event Structure Templates a. [x ACT

] (activity) b. [x <STATE>] (state)

c. [BECOME [x <STATE>]] (achievement)

d. [x CAUSE [BECOME [y <STATE>]]] (accomplishment)

e. [[x ACT

] CAUSE [BECOME [y <STATE>]]] (accomplishment)

(Rappaport Hovav & Levin 1998: 108) Resultatives are often analyzed with a complex event representation (Dowty 1979; Levin

& Rappaport Hovav 2005). The assumption is that verb meanings are complex and can be decomposed into multiple atomic events. On this analysis, resultatives involve a CAUSING SUBEVENT represented by the activity verb V, and a RESULT SUBEVENT, represented by the change of state, XP (in varied categories) (Rappaport Hovav & Levin 2001). The complex event structure is: [[x ACT

] CAUSE [BECOME [y

]]] (Levin 1999). Rapaport Hovav and Levin (2001: 783) put forward a series of constraints on the relation between two subevents:

(9) Constraints on two subevents

a. The subevents need not be temporally dependent.

b. The result subevent cannot begin before the causing subevent.

c. Only the result subevent can bound the event as a whole.

d. There is no intervening event between the causing subevent and the result subevent;

that is, causation is direct.

Lin (2004) summarizes the predicate decomposition literature. Beyond “Event structures”

(Rappaport Hovav & Levin 2001), we can find abundant literature, such as “Lexical Conceptual Structures” (Rappaport Hovav & Levin 1998), “Lexical Relational Structures”

(Hale 1993), “Semantic structures” (Pinker 1989), “Conceptual structures” (Jakendoff 1990), “Semantic Forms” (Van Valin & La Polla 1997), “L(exical)-syntax” (Travis 2002),

“I(nternal)-structure” (Zubizarreta & Oh 2007), and “First-phase syntax” (Ramchand 2008).

The event analysis in general sheds some light on understanding the internal relationship between the activity verb and the resultant predicate in this construction.

1.2 Semantic account of argument realization

Another prominent perspective of inquiry in resultatives is whether and how the properties of resultatives are derived from the argument structure of its component verbs, and how syntactic structures relate to argument structures. This will help us to know better about the internal relation between predicates and their arguments. A verb’s argument structure often specifies the lexical entry of each verb, determining how many numbers of arguments satisfied in syntax. Since Stowell (1981), argument structure has been portrayed as thematic-grids, which does not specify syntactic category or linear order, but only thematic roles (Marantz 1984).

Resultatives have been associated with the grammatical function of direct objects since

Simpson (1983). Simpson proposes that the resultant predicate (XP) must be predicated of

the syntactic object. The syntactic object includes surface objects, underlying objects, and

even fake reflexives (Simpson 1983: 142). Levin & Rappaport Hovav (1995) develop the generalization into the Direct Object Restriction, claiming that “a resultative phrase may be predicated of an immediately postverbal NP, but may not be predicated of a subject or of an oblique complement” (Levin & Rappaport Hovav 1995: 34), as in English in (10).

(10) a. John hammered the metal flat. (thematic object) b. John drank the teapot empty. (unselected object) c. John laughed himself silly. (fake reflexive)

d. The river froze it solid. (derived subject/underlying object)

(Levin & Rappaport Hovav 1995) However, Rappaport Hovav and Levin (2001) abandon their syntactic explanation on DOR later, since the subject-predicated resultant XPs are also found in English. In (11), the resultant phrase out of Bethlehem, clear of the rocks, across the room and across the park predicate the subject instead of the object of each sentence.

(11) a. The wise men followed the star out of Bethlehem.

b. The sailors managed to catch a breeze and ride it clear of the rocks.

c. John danced mazurkas across the room.

d. The children played leapfrog across the park. (Mateu 2005: 57) Mateu (2005) responds that the so-called validity of the DOR involves an adjunct XPs, claiming that the relevant problematic examples are underlying unaccusative constructions instead. In Dutch, the auxiliary selection in (12a) and the use of participle in prenominal position in (12b) show unaccusativity of these sentences.

(12) a. De politie is de dief tot zijn huis gevolgd (Dutch)

the police IS the thief to his house followed

b. deze mij tot aan de deur gevolgde politieman

this me until the door followed policeman (Mateu 2005:72)

Further supporting evidence is the impossibility of passivizing the example (13a) into (13b).

The impossibility is not because it is associated with a subject-predicated resultant XP, but rather that it is an unaccusative construction (Mateu 2005).

(13) a. The wise men followed the star out of Bethlehem.

b. *The star was followed out of Bethlehem. (Mateu 2005)

In contrast, the grammaticality of “The thief was followed to his house” is because the sentence has the additional reading within a transitive structure. The absence of the agent removes the DOR violation as in the passive construction. Mateu (2005) summarized into two sense contrasts:

(14) a. Correlated motion sense: Unaccusative structure (‘BE’ selected)

b. Detective-suspect/causative sense: Transitive structure (‘HAVE’ selected) The unaccusative use of follow is also shown in German (15a) and Dutch (15b).

(15) a. Die Polizei ist dem Dieb zu seinem Haus gefolgt (German) the police IS de thief-dat to his house followed

b. Die Polizei hat den Dieb zu seinem Haus verfolgt. (Dutch)

the police HAS de thief-acc to his house VER followed [Heiner Drenhaus]

(Mateu 2005: 73)

As noted, the unaccusative use of the verb follow is associated with BE-selection, with a

dative case assignment. The transitive use is represented with HAVE-selection, assigning

accusative case. The so-called counter-example to DOR actually does not challenge the restriction. Mateu claims that the adjunct analysis is the most natural approach for directional phrases, like those seemingly with subject-predicated resultant XPs.

1.3 Underlying syntactic representations

My major concern in this thesis is with the syntactic literature, and how the syntactic approach characterizes the resultative construction in general. Among current syntactic approaches, the small clause analysis (Hoekstra 1988, 1992) and the ternary branching analysis (Carrier & Randall 1992) are briefly introduced next. I will also illustrate some limitations of these two syntactic approaches.

1.3.1 Small clause analyses

Hoekstra (1988, 1992) analyzed the resultative construction with a small clause theory. The basic idea behind the small clause approach is that the second predicate is considered as a secondary predicate, which forms a small clause with its subject. Three small clause rules are summarized by Hoekstra (1988). The postverbal NPs in this approach are analyzed as affected objects, independently from the action mentioned by the main verb.

(16) Small Clause Rules

a. It adds a small clause complement to the verb;

b. It eliminates the internal arguments of the verb;

c. It gives the verb a causative reading. (Hoekstra 1988)

According to Hoekstra (1992), the predication must be stage-level, dynamic, and not inherently bounded. The sentences in (17) are grammatical, but problematic in (18). We can see the construction clashes when the predication is individual-level (18a), undynamic (18b), and bounded in nature (18c).

(17) a. They danced the night away.

b. Morris moaned his way out of the hall.

c. The dog barked the chickens awake.

(18) a. *This encyclopedist knows all books superfluous. (individual level) b. *Medusa saw the hero into stone. ([-dynamic])

c. *The psychopath killed the village into a ghost town. ([+bounded])

(Hoesktra 1992)

An important contribution of the small clause theory of resultatives is to show shortcomings of some competing lexicalist theories. As Hoekstra (1988: 138) noticed, “the common distinction between lexical word making and non-lexical sentence making is questionable at best”. For instance, these examples in (19) are identical, both “consisting of an activity denoting verb, taking a small clause complement which is interpreted as a resulting state”

(Hoekstra 1988: 138). The distinction between the syntactic object (19a) and the morphological object (19b) seems to be questionable, as generally assumed in lexicalist theories (Mateu 2001).

(19) a. dat Jan zich drunken drinkt. (syntactic object) that John himself drunk drinks

b. dat Jan zich bedrinkt. (morphological object)

that John himself BE-drinks

c. drinkt [SC zich {drunken/BE-}] (Mateu 2001)

However, the small clause theory does not emphasize the semantic distribution of the postverbal NP (i.e. the subject in small clause), or the resultative phrase (the predicate in small clause). Consequently, no proper semantic restriction guarantees an interpretable resultative phrase. Boas (2011) shows that the small clause analysis cannot specify the semantic selection restriction to postverbal constituents in resultatives.

(20) a. Pat ran his Nikes {threadbare/?blue/*new/*small}

b. Claire painted the door {red/?old/*visible/*broken}.

c. Chris drank {Martin/?himself/*his Martini/*the glass} under the table.

d. Nicole danced {Sacha/herself/?her cat/*her goldfish}tired. (Boas 2011:2)

The analysis also lacks empirical motivation to account for the postverbal constituent with lexically intransitive matrix verbs, such as run in Pat ran his Nikes threadbare. That is, if the activity verb and the postverbal argument are independent, the selected and unselected postverbal arguments make no difference in the small clause approach.

1.3.2 Ternary branching analyses

Carrier & Randall (1992) account for resultatives in a ternary analysis, dividing verb

phrases into: the activity verb, the NP and the resultative phrase. The assumption is that the

resultative phrase and the postverbal NP are both arguments of the activity verb.

(21) a. Claire [

painted [

the door] [

red]]

b. Pat ran his Nikes threadbare.

In the transitive resultative pattern (21a), the postverbal NP is the internal argument of the verb. The ternary branching analysis suggests that NP and AP are sisters of the activity verb, forming argument hood. According to Carrier & Randall, in the transitive resultative pattern (21a), it should mean as a consequence of painting the door, the door became red.

The interpretation supports that the door is an argument of the activity verb. They also note that in the intransitive resultative pattern (21b), the postverbal NP receives only one theta- role from the resultative phrase, but not from the ergative activity verb. They therefore exclude postverbal NPs of intransitive verbs from the set of arguments of the verb in the resultative construction.

Boas (2011) critiques this kind of analysis. Firstly, it cannot distinguish the selected and non-selected postverbal NPs in lexically transitive activity verbs: such as drink, wash and fry. This analysis cannot explain the semantic selection of postverbal NPs and resultative phrases. For instance, the postverbal object is not the semantically selected argument of the activity verb in (22a) and (22c), but selected in (22b).

(22) a. Melissa drank the teapot empty.

b. Dave washed the soap out of his eyes.

c. Ed fried the pan black. (Boas 2011)

Secondly, the exclusion of postverbal NPs out of arguments in intransitive verbs cannot explain why postverbal NPs exhibit the behavior of arguments under passivization (The Nikes were run threadbare) (Boas 2011).

So far both the small clause approach and the ternary analysis seem insufficient to explain the resultative construction. This leaves some space available to further investigate the syntactic derivation of this construction.

1.4 Resultative adverbs in generative grammar

In this section, an overall background of resultative adverbs is provided in generative grammar. Since adverb resultatives will be modeled in a unified explanation of broad resultative construction, it is necessary to know the general existing assumption. Geuder (2002) notes that resultatives in English can be coded with an adverb, rather than an adjective, preposition, or noun phrase. The term “adverb” refers to adverbial modifiers, which are morphologically derived from an adjectival base, or are formally identical to adjectives.

(23) a. They decorated the room beautifully.

b. She fixed the car perfectly.

c. She grows chrysanthemums marvelously.

(24) a. I opened the door wide.

b. I shut the door tight. (Geuder 2002:69)

The adverbs beautifully, perfectly, and marvelously in (23) imply respectively that the room was in a beautiful state as a result of the action of decorating it, that the car was perfect as a result of fixing it by her, and that the chrysanthemums grown by the subject referent were marvelous (Geuder 2002)

.

However, adverbs like wide and tight are analyzed in (24) as verbal modifiers in Parsons (1990) and Geuder (2002). The co-occurrence with an adjective ensures that wide and tight are adverbs, although they lack the suffix –ly, as we can see in “The door is wide open”.

The adverb occurs with the verb that already has its resultative state specified. For instance, the verb to open “takes its name from a predicate that describes a resultant state (i.e. the adjective open), so the slot of the result predicate is already filled” (Geuder 2002: 71). The word wide is a modifier and not the resultant state itself.

Parsons (1990) analyzes the oriented adverb as modification of the resultant state, and proposes the accomplishment as a series of three subevents in the event semantic framework. Parsons uses Culminate and Hold as predicates, which denote aspects. The decompositions of x open y and x open y wide are represented:

4

In addition to the resultative object-orientation, the adverb can also modify the activity event: the way of decorating (of the room) is beautiful, rather than the room is beautiful as a result of the decorating action. The alternative interpretation is supported by the non co-occurrence of this adverb and a regular result state. Thus, I propose that these two readings are derived from different syntactic structures. The object-orientation is generated when the resultative adverb is derived as a predicate; the modification is produced when the adverb is analysed as a manner modifier.

(iii) Arthur hammered the metal flat beautifully.

(25). x open y:

(Ǝe) [Cul (e) & Agent (e, x) & (Ǝe’) [Cul (e’) & Theme (e’, y) & CAUSE (e, e’) &

Ǝs [open (s) & Theme (s, y) & Hold (s) & BECOME (e’, s)]]]

(26). x open y wide:

(Ǝe) [Cul (e) & Agent (e, x) & (Ǝe’) [Cul (e’) & Theme (e’, y) & CAUSE (e, e’) &

Ǝs [open (s) & Theme (s, y) & Hold (s) & BECOME (e’, s) & Being-wide (s)]]]

(Parsons 1990)

Geuder (2002) agrees the modification analysis to adverbs, as argued in adjectival resultatives in (26). However, Geuder further explains that the result-orienting interpretation is realized in the compositional semantics. These adverbs are actually predicates of events in the compositional semantics. Ultimately they orient towards an individual,

created as a result of the event. The modification of this resultant individual is the source of their result-orienting interpretation.

Based on Geuder (2002), Levinson (2010) names this kind of adverbs pseudo-resultatives,

with puzzles for the syntax and semantics interface. What arguments are these modifiers modifying? Levinson shows pseudo-resultative modifiers are not DP-modifiers (modifiers of the predicate of individuals, denotes by the object DP), as in (27); not predicates of events, like manner adverbs in (28); and not implicit object modifiers, as argued by Geuder.

5

On Geuder’s account, this individual is not realized directly by any constituent in the syntax, but is assessed as part of the semantic contribution of the verb.

6

Levinson claims the resultative adverbs in Geuder (2002), are semantically similar to pseudo-

resultatives, but might differ from them in not bearing the –ly morpheme. That is, Levinson exclusively

focuses on adjective modifiers.